USE OF MULTIVARIATE SPATIAL ANALYSIS TO EVALUATE SOIL FERTILITY

Abstract

The spatial variability of soil attributes can be influenced by several factors such as soil formation (parent material, topography, vegetation, climate, and time), cultural practices (tillage, crop rotations, and fertilization), and erosion. Therefore, appropriate techniques are required to identify the variables that affect certain soil processes to improve soil management, liming and fertility practices in a farm. This study was done to demonstrate the applicability of multivariate geostatistics techniques to investigate the behavior of Phosphorus (P), Potassium (K) and Organic Matter (OM) in a 70 hectares experimental area in Araguari, Minas Gerais State, Brazil. These soil constituents were analyzed with the use of linear coregionalization model. Direct and cross p(p+1) variograms are adjusted, and then fractioned into principal components. The results showed a larger spatial influence of the variable P at micro and middle scale and shared influence of variables P and K at long scale. Due to the agronomical practices the influence of OM was small in the combined behavior of these constituents. P and K showed higher spatial dependence because of continuous applications of these fertilizers for soybean and corn cultivation. The results established a minimum limit of 320.18 meters for delimitating management zones for variable rate fertilizer zones.